There is a great need for improved non-invasive tools for evaluating breast cancer. A technique called magnetic resonance spectroscopy (MRS), generally performed along with a MRI examination, can produce information about the quantities of specific metabolites within a distinct lesion in a breast. The amount of signal produced by choline-containing compounds (tCho) is known to be higher in malignant tumors than in benign and normal tissues. Previous work by a number of groups has demonstrated that measuring tCho can provide clinically useful information. On clinical 1.5 T MR scanners, simply detecting a tCho signal indicates that a suspect lesion is malignant. It has also been shown that the tCho signal decreases in response to successful chemotherapy treatment, suggesting that it can be used as an early indicator of response. These findings are encouraging and demonstrate clinical potential. However, if breast MRS is to be made into a clinically usable tool, it will be necessary to produce quantitative, reliable measurements of tCho, rather than qualitative judgments of detectability. This is particularly critical for performing longitudinal studies, such as monitoring treatment response. Quantitative measurements are also required for performing diagnosis with higher-field scanners, as the increased sensitivity makes tCho detectable in benign lesions and normal breast tissues. The goal of this project is to develop and demonstrate a reliable methodology for making quantitative tCho measurements in breast lesions on clinical 3T MR scanners. The specific aims are to 1) optimize pulse sequences and aquisition methods, 2) evaluate reproducibility and identify sources of error, and 3) demonstrate feasibility of monitoring response to treatment in a small patient study. This translational research project will bring advanced research methods to a clinical MR system, and provide the foundation for large-scale, multi-center trials that are required to evaluate clinical efficacy.